Marco Dinarelli with his first journal publication in a IEEE review Marco Dinarelli
Web site of Marco Dinarelli in English  Site web de Marco Dinarelli en français  Sito web di Marco Dinarelli in italiano 


LIG (UMR 5217)
Office 327
700 avenue Centrale
Campus de Saint-Martin-d’Hères, France


Email:
marco [dot] dinarelli [at] univ-grenoble-alpes [dot] fr
marco [dot] dinarelli [at] gmail [dot] com

                        Curriculum Vitae           Profile of Marco Dinarelli in LinkedIn


Latest news

2023 / 10 / 07:
Paper accepted at the conference WMT 2023

2023 / 05 / 05:
Paper accepted at the conference LDK 2023

2023 / 03 / 18:
Paper accepted at the workshop on insights from negative results in NLP Insights 2023

Seq2Biseq - Bidirectional Output-wise Recurrent Neural Networks for Sequence Modelling

Content index:

Description

Seq2Biseq tool is the software used for the paper Seq2Biseq: Bidirectional Output-wise Recurrent Neural Networks for Sequence Modelling. It replaces, extends and improves the previous tool LD-RNN, used for the paper Label-Dependencies Aware Recurrent Neural Networks.
Seq2Biseq is coded in pytorch and it follows the same research trend as our previous papers, where a bidirectional output-side context is used for current decision. A schema of the high-level architecture is shown in the following image.
Seq2Biseq model architecture


The idea is similar to those used in Deliberation Networks, and Asynchronous bidirectional networks for Machine Translation.

Features

  • Bidirectional backward-forward decoding

Download

Please send me an email @univ-grenoble-alpes.

Licence

Seq2Biseq is provided under Creative-Commons BY-SA licence

Installation and usage

See the README file in the package.